Customer Success Management and Predictive Analytics

I have been working on CSM for quite some time now and thought of sharing my insights. So lets start with the basics, what is Customer Success Management?

The emerging role of Customer Success Management is about a solution to the core issues of customer portfolio development, retention and expansion. The concepts are similar to customer analytics (churn, up-sell/cross-sell, Segmentation ) in the space of telecom and BFSI. These concepts are applied to the the recurring revenue business of the SaaS world and we get a use-case very well suited to gain benefits from them. The cost of acquiring a new customer is pretty large and to cover that cost and to gain profit from a particular customer, it is imperative that a customer is retained for the longest possible time and opportunity of  up-sell/cross-sell converted.

Customer Success Management is different from traditional customer analytics, in that it focuses on a 360 degree view of the customer and a more engaged relationship in which you help your customer better understand their customer base and help them derive success from the product offered. It’s a far cry from the previous approach of just viewing customers as mere revenue stream.

Customer Success Management entails combining data from various different data sources, analyzing the data to understand the signature of healthy and at-risk customer, providing alerts and workflows to retain customers at-risk or up-sell/cross-sell to a healthy customer. It combines the world to BI reporting and predictive analytics workflows over a combined data store to provide a 360 degree of a customer.

From the predictive analytics point of view most of the SaaS organization’s data provides a small n large p problem. The number of data sources and the features that can be derived are large but the training rows are smaller in size. Appropriate modeling technique has to be used to address this scenario. Also as the focus is to understand the customer base, white box modeling approach play a very important role.

The granular level usage data could be pretty large and hence platform like Hadoop to preprocess and aggregate data into a feature set is an integral part of the tech stack. As scaling up the analytics is of utmost importance if CSM predictive analytics is provided as a SaaS offering, early focus on automation of the Predictive Analytics work-flow is of paramount importance.